Hello, DSP!

This is a self-contained piece of C++ audio code. The interesting bit is main.cpp.

It generates a 2-second 440Hz sine wav, and writes it as a WAV file.

To build and run

Using the Makefile:

make

This builds main.cpp into out/main, and then runs it from the out/ directory, producing out/tone-440.wav.

wav.h

This is a simple library to read and write 16-bit WAV files - enough for demos, experimenting and learning, but probably not serious use.

Shape and sample-rate

The basic properties are .sampleRate, .channel and .length():

Wav wav;
// Sample-rate, as a double
wav.sampleRate = 48000;
// Channel-count as an unsigned int
wav.channels = 2;

int length = wav.length(); // length (for each channel)
wav.resize(2, 1000); // resize with channel/length

WAV sample-rates are actually 32-bit integers, but .sampleRate is a double since we often divide by it.

Sample data

The most convenient way to read/write samples is wav[channel][index]. Samples are scaled to the unit range (-1, 1), and converted on input/output.

The actual data is stored in .samples, which is a std::vector<double>:

wav.samples.resize(1000);
wav.samples[0] = 0;

Multichannel WAV data is interleaved, so for a stereo file:

• wav.samples[0] is the first left, sample
• wav.samples[1] is the first right sample
• wav.samples[2] is the second left sample
• etc.

Reading and writing

You read/write with .read() and .write(), using filenames as std::string.

These return a result object, which converts to bool for success/failure, and has a .reason string for human-readable errors.

Wav wav;
auto result = wav.read("input.wav");
if (!result) {
	std::cerr << "WAV error: " << result.reason << "\n";
}

The latest result is also stored as wav.result.

Since logging an error is so common, you can do this with the .warn() method:

if (!wav.write("output.wav").warn()) {
	// handle failure
}

License

It's released as CC0. If you need anything else, contact me.